Postal Rate Commission
Submitted 9/24/01 1:10 pm
USPS-T-2
BEFORE THE
POSTAL RATE COMMISSION
WASHINGTON, D.C. 20268-0001
---------------------------------------------------------:
Postal Rate and Fee Changes, 2001
:
:
---------------------------------------------------------:
DIRECT TESTIMONY OF
JENNIFER J. XIE
ON BEHALF OF THE
UNITED STATES POSTAL SERVICE
Docket No. R2001-1
TABLE OF CONTENTS
TABLE OF CONTENTS ................................................................................................... i
AUTOBIOGRAPHICAL SKETCH .................................................................................... 1
PURPOSE AND SCOPE................................................................................................. 2
I.
HIGHWAY ................................................................................................................ 3
II.
FREIGHT RAIL ........................................................................................................ 6
III. PASSENGER RAIL (AMTRAK)................................................................................ 8
IV. COMMERCIAL AIR................................................................................................ 11
V.
NETWORK AIR...................................................................................................... 14
APPENDIX .................................................................................................................... 17
Table 1. BY00 Inter-BMC Highway Costs and Confidence Intervals ......................... 18
Table 2. BY00 Intra-BMC Highway Costs and Confidence Intervals ......................... 19
Table 3. BY00 Inter-SCF Highway Costs and Confidence Intervals .......................... 20
Table 4. BY00 Intra-SCF Highway Costs and Confidence Intervals .......................... 21
Table 5. BY00 Freight Rail Costs and Confidence Intervals ...................................... 22
Table 6. BY00 Passenger Rail (Amtrak) Costs and Confidence Intervals ................. 23
Table 7. BY00 Amtrak - Roadrailer Distribution Key and Costs................................. 24
Table 8. BY00 Commercial Air Costs and Confidence Intervals................................ 25
Table 9. BY00 Eagle Network Costs and Confidence Intervals ................................. 26
Table 10. BY00 Western Network Costs and Confidence Intervals........................... 27
i
1
1
2
3
DIRECT TESTIMONY
OF
JENNIFER J. XIE
4
5
AUTOBIOGRAPHICAL SKETCH
6
My name is Jennifer J. Xie. I am a Mathematical Statistician in Cost
7
Systems, Finance. Since joining the Postal Service in 1995, I have worked on
8
statistical design and estimation issues for the Transportation Cost System,
9
Origin-Destination Information System, the Revenue, Pieces and Weight System,
10
and the System for International Revenue and Volume Outbound. I have served
11
as project manager for the Transportation Cost System since August 1997.
12
Prior to joining the Postal Service, I was employed as a senior Load
13
Research Analyst by the Potomac Electric Power Company (PEPCO). My
14
primary responsibility at PEPCO was to conduct statistical analyses using
15
customer billing and survey information to quantify energy reductions attributable
16
to various conservation programs. I also provided technical support in several
17
PEPCO rate case proceedings.
18
I received a B.S. in Electrical Engineering from Jiangsu Institute of
19
Technology, China, in 1982 and a M.S. in System Engineering from Hohai
20
University, China, in 1987. I earned a Ph.D. in Operations Research and Applied
21
Statistics from George Mason University in 1992. I am a member of the
22
American Statistical Association and the Washington Statistical Society.
2
1
2
PURPOSE AND SCOPE
3
The purpose of my testimony is to describe the Transportation Cost
4
System (TRACS), which is a statistical information system used to distribute
5
base year (BY) purchased transportation costs to mail categories. TRACS is
6
composed of five distinct subsystems: Highway, Freight Rail, Passenger Rail
7
(Amtrak), Commercial Air and Network Air. All five subsystems are continuous,
8
ongoing survey systems designed to collect data from the different transportation
9
modes. Each subsystem employs its own survey design and estimation
10
procedures. My testimony covers the general design for each subsystem, the
11
types of estimates each produces, and tables of major estimates and confidence
12
limits.
13
3
1
I.
HIGHWAY
2
The TRACS-Highway subsystem produces separate distribution keys for
3
four types of purchased highway contracts: Inter-BMC1, Intra-BMC, Inter-SCF2,
4
and Intra-SCF. The cost for each contract type is composed of one or more
5
accounts. The universe under study is all mail moved on contracts whose costs
6
accrue to these highway accounts:
7
Inter-BMC:
Account Number 53131 (regular Inter-BMC);
8
Intra-BMC:
Account Number 53127 (regular Intra-BMC);
9
Inter-SCF:
Account Numbers 53124 (regular Inter-SCF), 53609 (regular
10
inter-P&DC), 53614 (regular inter-cluster), and 53618
11
(regular inter-area);
12
Intra-SCF:
13
Account Numbers 53121 (regular intra-SCF), 53601 (regular
intra-P&DC), and 53605 (regular intra-district).
14
The primary sampling unit (PSU) for all four contract types is the route-
15
trip-stop-day, which is defined as all mail unloaded from a truck at one facility on
16
a specific trip, on a specific day. The survey design is essentially the same for all
17
the contract types, though each of them has its own sampling frame. Each
18
highway sampling frame is a list of stop-days3. There are three major steps
19
involved in constructing the sampling frames. In the first step, routing and
20
operation information is extracted from the National Air and Surface System
1
Bulk Mail Center.
Sectional Center Facility.
3
Abbreviated name for route-trip-stop-days
2
4
1
(NASS)4 for all the highway contract routes that are expected to be in operation
2
in the upcoming quarter. The information extracted from NASS includes the
3
route number, the trip number, the facilities where the vehicle stops, and the
4
days of a week when the trip operates. In the second step, account information
5
is extracted from the Highway Pay Master File5 for the same contract routes.
6
The account information is used to group the contracts into the four contract
7
types. In the third and final step, the ZIP Code and facility type are extracted
8
from the NASS Facility File6 for each stop on a route. The facility information is
9
used for stratifying the sampling frame, as well as for administering the survey.
10
The sample design consists of three stages. In the first stage, within each
11
contract type, the stop-days are stratified based on the facility type of the stop
12
and whether the trip is inbound or outbound. A systematic random sample of
13
stop-days is selected from each stratum. In the second stage, for each selected
14
stop-day, a subsample of wheeled containers, pallets and loose items7 off-loaded
15
at the test facility is selected. From selected containers, a third stage sample of
16
items is selected. For pallets and loose items selected at the second stage,
17
there is no third stage sample. All selected mail is recorded.
18
Weight and volume information by mail category is recorded for the
19
contents of sampled items. For sampled pallets, the dimensions of the pallet and
4
See Section III and Appendix I-1 of TRACS Highway Subsystem Statistical and Computer
Documentation, filed as USPS-LR-J-32, for additional details regarding NASS.
5
See Section III and Appendix I-2 of TRACS Highway Subsystem Statistical and Computer
Documentation, filed as USPS-LR-J-32, for additional details regarding Pay Master File.
6
See Section III and Appendix I-3 of TRACS Highway Subsystem Statistical and Computer
Documentation, filed as USPS-LR-J-32, for additional details regarding NASS Facility File.
7
Items include pieces, parcels, bundles, sacks, trays, or tubs. Items that are not in wheeled
containers or on pallets are called loose items.
5
1
the proportion of the pallet's space occupied by each mail category are recorded.
2
In order to develop estimates of cubic-foot-miles, data collectors also record the
3
facility where the item, or the pallet, was loaded onto the vehicle (to establish
4
miles traveled) and the percent of vehicle floor occupied by palletized mail,
5
containerized items, and loose items (to establish cubic-feet utilized). Data are
6
recorded directly into portable microcomputers using the Computerized On-Site
7
Data Entry System (CODES) software. From the sample data, the cubic-foot-
8
miles transported for each contract type are estimated by mail category.
9
Distribution keys are calculated by dividing the mail category cubic-foot-
10
miles by the total cubic-foot-miles. Separate distribution keys are calculated for
11
each quarter for each of the four contract types, and are used to distribute
12
quarterly costs by contract types. Annual costs, shown in Tables 1-4, are the
13
sums of the quarterly costs. The confidence intervals of annual costs, also
14
shown in Tables 1-4, are derived from the coefficients of variation (CVs) of the
15
quarterly distribution keys.
16
A more detailed description of the TRACS-Highway sample design and
17
estimation methodology is contained in Sections I-VII of Library Reference
18
USPS-LR-J-32, TRACS Highway Subsystem Statistical and Computer
19
Documentation. TRACS data collection procedures are detailed further in
20
Chapter 5 of Handbook F-65, filed as Library Reference USPS-LR-J-14, with
21
supplemental instructions in Library Reference USPS-LR-J-34, Supplemental
22
Statistical Programs Policies and Data Collection Instructions. The CODES-
6
1
TRACS software, used on laptop computers to record the data, is documented in
2
Section 1 of Library Reference USPS-LR-J-35.
3
II.
FREIGHT RAIL
4
The TRACS-Rail subsystem produces distribution keys for the Inter-BMC
5
freight rail account (53143). The universe under study is all mail transported via
6
freight rail whose costs accrue to this account. The PSU for freight rail is the
7
origin-destination-day, which is defined as all mail being transported from a given
8
origin BMC to a given destination on a given day. While highway transportation
9
is contracted by route, with one route consisting of multiple trips and stops,
10
freight rail is contracted between one origin and one destination. A trip between
11
the origin and the destination facility is referred to as a rail movement. The
12
freight rail sampling frame is a list of all origin-destination-days (movements) with
13
a BMC origin which are not destined for a mail bag depository or mailer’s plant,
14
and are not used exclusively for empty equipment. The freight rail sampling
15
frame is developed using 12 weeks of historical records from the Rail
16
Management Information System (RMIS)8. The information extracted from RMIS
17
includes the origin BMC, the destination facility, the date of arrival, the number of
18
vans on the movement, and the cost of the movement.
19
The sample design consists of four stages. In the first stage, a random
20
sample of rail movements is selected from the sampling frame. In the second
21
stage, one van is randomly selected from the vans on the selected movement. In
8
See Section II-1 and Appendix I-2 of TRACS Freight Rail Subsystem Statistical and Computer
Documentation, filed as USPS-LR-J-33, for additional details regarding RMIS.
7
1
the third stage, a subsample of wheeled containers, pallets and loose items off-
2
loaded from the test van is selected. From selected containers, a fourth stage
3
sample of items is selected. For pallets and loose items selected at the third
4
stage, there is no fourth stage sample. All selected mail is recorded. The freight
5
rail sample design at the third and fourth stages is the same as the highway
6
sample design at the second and third stages.
7
Weight and volume information by rate category is recorded for the
8
contents of sampled items. For sampled pallets, the dimensions of the pallet and
9
the proportion of the pallet's space occupied by each mail category are recorded.
10
Data collectors also record the percent of van floor space occupied by palletized
11
mail, containerized items, and loose items. Data are recorded directly into
12
portable microcomputers using CODES software.
13
The sample data are expanded, by mail category, to the cubic-foot-miles
14
of the test van. The cost for the trip is multiplied by the cubic-foot-mile
15
proportions to estimate mail category costs for the trip. The costs for tested trips
16
are then expanded to represent all trips in the quarter.
17
Distribution keys are calculated by dividing the expanded costs for a mail
18
category by the total expanded costs. Separate distribution keys are calculated
19
for each quarter, and are used to distribute quarterly costs. Annual costs, shown
20
in Table 5, are the sum of quarterly costs. The confidence intervals for annual
21
costs, also shown in Table 5, are derived from the CVs of the quarterly
22
distribution keys.
8
1
A more detailed description of the TRACS-Rail sample design and
2
estimation methodology is contained in Sections I-V of Library Reference USPS-
3
LR-J-33, TRACS Freight Rail Subsystem Statistical and Computer
4
Documentation. TRACS data collection procedures are detailed further in
5
Chapter 5 of Handbook F-65, filed as Library Reference USPS-LR-J-14, with
6
supplemental instructions in Library Reference USPS-LR-J-34, Supplemental
7
Statistical Programs Policies and Data Collection Instructions. The CODES-
8
TRACS software, used on laptop computers to record the data, is documented in
9
Section 1 of Library Reference USPS-LR-J-35.
10
11
12
III.
PASSENGER RAIL (AMTRAK)
The TRACS-Amtrak subsystem produces distribution keys for the
13
passenger rail service account, 53142. The universe under study is all mail
14
transported via Amtrak whose costs accrue to this account. In fiscal year 2000,
15
the Postal Service contracted for mail transportation on 39 Amtrak trains. Each
16
of them makes multiple trips a week along the same route from a specific origin,
17
stopping at various points en route, to a final destination.
18
The PSU is a trip, which consists of all mail unloaded from an Amtrak train
19
at all the stops along a specific trip. A trip is uniquely determined by its train
20
number and the day the train departs. It can extend over more than one
21
calendar date. The sampling frame is a list of trips. It is developed by extracting
22
the routing and operation information for all Amtrak trains in the NASS database.
23
Information extracted from NASS includes the train number, the frequency (days
9
1
of a week when the train departs), the origin, cities at which it stops, and the
2
arrival time for each stop. The 39 trains are divided into two groups (low cost
3
and high cost) based on their contract costs obtained from the Amtrak Train Cost
4
File9.
5
The sample design consists of four stages. At the first stage, trips are
6
stratified by train; one trip per quarter is selected from each low cost train and
7
two trips from each high cost train. All stops on the selected trip are sampled. At
8
the second stage, one car is randomly selected from all cars offloading mail at
9
each stop of the sampled trip. At the third stage, a subsample of wheeled
10
containers, pallets and loose items is selected from the mail off-loaded from the
11
selected car. From selected containers, a fourth stage sample of items is
12
selected. There is no fourth stage sample for pallets and loose items selected at
13
the second stage. All selected mail is recorded. The Amtrak sample design at
14
the third and fourth stages is essentially the same as the Highway sample
15
design.
16
Weight and volume information by rate category is recorded for the
17
contents of sampled items. For sampled pallets, the dimensions of the pallet and
18
the proportion of the pallet's space occupied by each mail category are recorded.
19
To permit estimation of square-foot-miles, data collectors also record the facility
20
where the item, or the pallet, was loaded onto the train (to establish the miles
21
traveled) and the total number of containers and loose items off-loaded from the
9
See Section I and Appendix I of TRACS Passenger Rail (Amtrak) Subsystem Statistical and
Computer Documentation, filed as USPS-LR-J-30, for additional details regarding Amtrak Train
Cost File.
10
1
entire train (to establish the square-feet utilized). Data are recorded directly into
2
portable microcomputers using CODES software.
3
The sample data are expanded, by mail category, to the square-foot-miles
4
of the sampled trip. Mail category proportions of square-foot-miles are calculated
5
for each sampled trip and weighted by the trip cost. The weighted mail category
6
proportion is multiplied by the train cost to estimate the mail category cost for the
7
train. The distribution key is calculated by dividing the sum of mail category
8
costs for the 39 trains by the total cost for the 39 trains. Separate distribution
9
keys are calculated for each quarter, and are used to distribute quarterly costs.
10
Annual costs, shown in Table 6, are the sum of quarterly costs. The confidence
11
intervals for annual costs, also shown in Table 6, are derived from the CVs of the
12
quarterly costs.
13
A more detailed description of the TRACS-Amtrak sample design and
14
estimation methodology is contained in Sections I-VII of Library Reference
15
USPS-LR-J-30, TRACS Passenger Rail (Amtrak) Subsystem Statistical and
16
Computer Documentation. TRACS data collection procedures are detailed
17
further in Chapter 5 of Handbook F-65, filed as Library Reference USPS-LR-J-
18
14, with supplemental instructions in Library Reference USPS-LR-J-34,
19
Supplemental Statistical Programs Policies and Data Collection Instructions. The
20
CODES-TRACS software, used on laptop computers to record the data, is
21
documented in Section 1 of Library Reference USPS-LR-J-35.
22
23
Once a year, TRACS samples mail transported on Amtrak-Roadrailer. A
distribution key is developed to allocate Amtrak-Roadrailer costs. In BY00, the
11
1
Roadrailer sample was conducted during the period of June 17 – July 7. Each
2
Roadrailer route was treated as a regular train. Every trip was treated as a PSU.
3
Four tests were conducted for each Roadrailer. To the extent possible, the tests
4
were taken on different days of the week to help ensure a representative mix
5
across days. At all stops of each trip, data were collected from unloaded mail
6
using normal Amtrak test procedures. Data were recorded using the CODES
7
software for Amtrak tests and transmitted to the mainframe computer along with
8
Amtrak test data. The Amtrak expansion procedure was used in the Roadrailer
9
expansion process. Each Roadrailer was treated as if it was a train and the four
10
tested trips were treated as PSU’s. A distribution key was calculated using the
11
expanded Roadrailer sample, and was used to distribute the BY00 Roadrailer
12
costs. Results are shown in Table 7.
13
A more detailed description of the Roadrailer sample design and
14
estimation methodology for the BY00 study is provided in USPS-LR-I-
15
433/R2000-1. The expansion programs, inputs to the expansion programs and
16
SAS logs used to generate the distribution key are contained in Appendix V: Lists
17
of Source Code and Data on CD-ROM, of USPS-LR-J-30.
18
IV.
19
COMMERCIAL AIR
The TRACS-Air subsystem produces distribution keys for commercial air
20
transportation costs. The universe under study is all mail transported under
21
purchased transportation contracts on passenger airlines from a domestic origin
22
to a domestic destination. It does not include mail traveling via air taxi or the
23
Eagle, Western or Christmas Networks. The PSU is a flight-day, and is defined
12
1
as all mail being dispatched from the specified origin on a given day via a
2
particular airline and flight with the same first-leg destination reflected on the
3
routing label. The sampling frame is constructed by extracting the recent
4
dispatch records from the Air Contract Support System (ACSS)10. ACSS
5
maintains the routing information (carrier, origin, destination, and date of the
6
flight) about an individual dispatch, and specifies the gross weight and primary
7
mail class for all items in the dispatch. Flights whose schedules continue through
8
the upcoming postal quarter according to the Official Airline Guide (OAG) are
9
eligible for sampling.
10
The sample design consists of two stages. In the first stage, a random
11
sample of flight-days is selected from the sampling frame. In the second stage,
12
for each selected flight-day, a subsample of mail items dispatched for that flight is
13
selected for detailed sampling.
14
Dispatch information is recorded for each selected item. Weight and
15
volume information, by mail category, is then recorded for the contents of the
16
item. The dispatch information is used in the expansion process to link the
17
sample data with dispatch records in the Plan vs. Actual (PVA) file11. Similar to
18
the ACSS, but at a less aggregated mail class level, the PVA file maintains
19
individual dispatch information for all items transported by air during the entire
20
quarter. Data are recorded directly into a portable microcomputer using CODES
21
software.
10
See Section III and Appendix I-B of TRACS Commercial Air Subsystem Statistical and
Computer Documentation, filed as USPS-LR-J-29, for additional details regarding ACSS file.
11
See Section III and Appendix I-K of TRACS Commercial Air Subsystem Statistical and
Computer Documentation, filed as USPS-LR-J-29, for additional details regarding PVA file.
13
1
The sample data are expanded, by rate category and by primary mail
2
class, to the pound-miles of mail on the test flight and then to the total pound-
3
miles of mail for the quarter. PVA records are used to determine the pound-miles
4
of mail for the test flight, as well as the total pound-miles for the quarter for each
5
primary mail class. The expanded pound-miles for a mail category are obtained
6
by adding fully expanded pound-miles across all primary mail classes.
7
Distribution keys are calculated by dividing the expanded pound-miles for
8
a mail category by the total expanded pound-miles. Separate distribution keys
9
are calculated for each quarter, and are used to distribute quarterly costs.
10
Annual costs, shown in Table 8, are the sum of the quarterly costs. The
11
confidence intervals for annual costs, also shown in Table 8, are derived from the
12
CVs of the quarterly costs.
13
A more detailed description of the TRACS-Air sample design and
14
estimation methodology is contained in Sections I-VII of Library Reference
15
USPS-LR-J-29, TRACS Commercial Air Subsystem Statistical and Computer
16
Documentation. TRACS data collection procedures are detailed further in
17
Chapter 5 of Handbook F-65, filed as Library Reference USPS-LR-J-14, with
18
supplemental instructions in Library Reference USPS-LR-J-34, Supplemental
19
Statistical Programs Policies and Data Collection Instructions. The CODES-
20
TRACS software, used on laptop computers to record the data, is documented in
21
Section 1 of Library Reference USPS-LR-J-35.
22
14
1
V.
2
NETWORK AIR
The TRACS-Network Air subsystem produces two sets of distribution
3
keys, one for the Eagle Network and one for the Western Network. The universe
4
under study is all mail transported on the Eagle and Western Networks whose
5
costs accrue to the following accounts:
6
53541: Eagle Network Line Haul.
7
53543: Eagle Network Terminal Handling.
8
53545: Western Network Line Haul.
9
53546: Western Network Terminal Handling.
10
53547: Eagle Network Fuel.
11
It does not include mail transported on passenger airlines, day net, air taxi
12
or the Christmas Networks. Unlike commercial airline flights, network flights are
13
dedicated fully to transporting mail between the hub and one or more cities in the
14
network. The hub for the Eagle Network is Indianapolis, Indiana; for the Western
15
Network, it is Oakland, California. The sample design for each network is the
16
same.
17
The PSU is a city-day, which consists of all mail dispatched via network
18
flights from that city to the hub (as the first leg flight) on the specified day. The
19
sampling frame is a list of such city-days, and is constructed by extracting the
20
network flight data from the NASS database. For each network city, NASS
21
specifies the flight number, hub destination, flight frequency (days of a week
22
when the flight runs), and the closeout time. Each city is sampled three times per
23
quarter.
15
1
The sample design consists of two stages. In the first stage, for each city
2
in the network, three days are randomly selected for sampling. In the second
3
stage, a subsample of mail items dispatched via the network flight to the hub is
4
selected for detailed sampling.
5
Dispatch information is recorded for each selected item. Weight and
6
volume information, by rate category, are then recorded for the contents of the
7
item. The dispatch information is used in the expansion process to link the
8
sample data with dispatch records in the PVA file. Data are recorded directly into
9
a portable microcomputer using CODES software.
10
The sample data are expanded, by rate category and by primary mail
11
class, to the pound-miles of mail on the test flight and then to the total pound-
12
miles of mail transported on the network during the quarter. PVA records are
13
used to determine the pound-miles of mail for the test flight, as well as the total
14
pound-miles for the quarter for each primary mail class. The expanded pound-
15
miles for a mail category are obtained by adding fully expanded pound-miles
16
across all primary mail classes.
17
Distribution keys are calculated by dividing the expanded pound-miles for
18
a mail category by the total expanded pound-miles. Separate distribution keys
19
are calculated for each network for each quarter, and are used to distribute
20
quarterly non-premium costs. Annual costs, shown in Tables 9-10, are the sum
21
of the quarterly costs. The confidence intervals for annual costs, also shown in
22
Tables 8-9, are derived from the CVs of the quarterly costs.
16
1
A more detailed description of the TRACS-Network Air sample design and
2
estimation methodology is contained in Sections I-VII of Library Reference
3
USPS-LR-J-31, TRACS Network Air Subsystem Statistical and Computer
4
Documentation. TRACS data collection procedures are detailed further in
5
Chapter 5 of Handbook F-65, filed as Library Reference USPS-LR-J-14, with
6
supplemental instructions in Library Reference USPS-LR-J-34, Supplemental
7
Statistical Programs Policies and Data Collection Instructions. The CODES-
8
TRACS software, used on laptop computers to record the data, is documented in
9
Section 1 of Library Reference USPS-LR-J-35.
17
APPENDIX
Table 1. BY00 Inter-BMC Highway Costs and Confidence Intervals
Table 2. BY00 Intra-BMC Highway Costs and Confidence Intervals
Table 3. BY00 Inter-SCF Highway Costs and Confidence Intervals
Table 4. BY00 Intra-SCF Highway Costs and Confidence Intervals
Table 5. BY00 Freight Rail Costs and Confidence Intervals
Table 6. BY00 Passenger Rail (Amtrak) Costs and Confidence Intervals
Table 7. BY00 Amtrak - Roadrailer Distribution Key and Costs
Table 8. BY00 Commercial Air Costs and Confidence Intervals
Table 9. BY00 Eagle Network Costs and Confidence Intervals
Table 10. BY00 Western Network Costs and Confidence Intervals
18
Table 1. BY00 Inter-BMC Highway Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.129
0.159
0.370
0.491
0.132
0.342
0.054
0.172
0.101
0.055
0.081
0.069
0.390
0.433
0.239
Lower
95% C.L.
($1,000)
8,165
5,291
24
8
6,203
45
34,873
4,422
69,309
55,130
14,008
20,572
183
105
1,965
Upper
Cost
95% C.L.
($1,000)
($1,000)
10,929
13,694
7,682
10,072
88
151
229
449
8,375
10,547
135
226
38,989
43,104
6,661
8,900
86,486
103,663
61,851
68,572
16,664
19,319
23,802
27,033
775
1,366
696
1,287
3,700
5,434
267,060
19
Table 2. BY00 Intra-BMC Highway Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.097
0.133
0.268
0.274
0.107
0.282
0.076
0.190
0.117
0.050
0.102
0.090
0.415
0.335
0.310
Lower
95% C.L.
($1,000)
12,548
7,364
69
32
16,046
378
20,881
6,789
45,651
85,511
19,089
13,826
44
87
1,065
Cost
($1,000)
15,506
9,972
145
69
20,306
846
24,546
10,816
59,215
94,753
23,869
16,792
234
254
2,714
280,037
Upper
95% C.L.
($1,000)
18,464
12,581
221
106
24,567
1,314
28,210
14,844
72,779
103,995
28,649
19,758
423
421
4,363
20
Table 3. BY00 Inter-SCF Highway Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.055
0.158
0.205
0.316
0.081
0.220
0.129
0.339
0.211
0.162
0.193
0.251
0.611
0.343
0.457
Lower
95% C.L.
($1,000)
136,904
44,430
1,308
689
97,320
3,477
37,848
996
22,171
12,239
2,495
3,334
344
849
Cost
($1,000)
153,508
64,385
2,187
1,804
115,561
6,117
50,600
2,968
37,771
17,929
4,019
6,566
703
1,050
8,159
473,327
Upper
95% C.L.
($1,000)
170,112
84,340
3,066
2,920
133,801
8,756
63,352
4,939
53,371
23,620
5,542
9,798
1,546
1,756
15,469
21
Table 4. BY00 Intra-SCF Highway Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.046
0.111
0.175
0.210
0.063
0.253
0.081
0.232
0.133
0.097
0.136
0.153
0.284
0.256
0.314
Lower
95% C.L.
($1,000)
104,087
31,812
1,057
417
126,776
4,567
57,359
12,642
68,309
46,786
10,048
5,454
445
501
1,703
Cost
($1,000)
114,480
40,687
1,607
707
144,506
9,057
68,182
23,222
92,308
57,794
13,688
7,801
1,001
1,007
4,433
580,483
Upper
95% C.L.
($1,000)
124,872
49,561
2,158
998
162,237
13,548
79,005
33,802
116,307
68,803
17,329
10,148
1,558
1,512
7,164
22
Table 5. BY00 Freight Rail Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
18.469
10.791
311.531
809.714
20.369
284.422
0.182
0.124
0.010
0.078
0.124
0.091
28.556
81.693
0.793
Lower
95% C.L.
($1,000)
8,328
5,577
58,664
19,854
6,363
9,825
-
Cost
($1,000)
468
570
23
8
620
48
12,939
7,360
59,857
23,450
8,416
11,961
291
135
3,944
130,089
Upper
95% C.L.
($1,000)
17,424
12,634
14,169
12,740
25,372
26,534
17,550
9,142
61,049
27,047
10,469
14,097
16,598
21,677
10,070
23
Table 6. BY00 Passenger Rail (Amtrak) Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.335
0.377
0.360
0.971
0.230
0.000
0.043
0.617
0.153
0.384
0.091
0.751
2.375
0.179
0.621
Lower
95% C.L.
($1,000)
1,044
1,745
8
1,220
72,306
1,179
199
24
256
-
Cost
($1,000)
3,031
6,662
29
231
2,224
78,931
75
1,685
804
30
142
603
394
406
95,246
Upper
95% C.L.
($1,000)
5,019
11,580
49
670
3,229
85,556
165
2,191
1,409
35
350
3,409
533
899
24
Table 7. BY00 Amtrak - Roadrailer Distribution Key and Costs
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
Distribution
Key
.2832
.2913
.0109
.0025
.0152
.3383
.0002
.0189
.0048
.0047
.0166
.0111
.0016
.0006
1.0000
Cost
($1,000)
1,507
1,550
58
13
81
1,800
1
100
26
25
88
59
8
3
5,321
25
Table 8. BY00 Commercial Air Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.102
0.122
0.335
0.354
0.140
0.223
0.222
0.399
0.202
0.789
0.424
0.266
0.473
0.639
0.232
Lower
95% C.L.
($1,000)
166,019
133,272
1,849
540
229,434
5,985
8,549
54
11,034
364
757
83
10,984
Upper
Cost
95% C.L.
($1,000)
($1,000)
207,567
249,114
175,228
217,183
5,385
8,921
1,761
2,982
315,962
402,489
10,627
15,269
15,154
21,759
247
441
18,275
25,517
2,274
5,788
2,149
3,934
1,584
2,410
1,151
2,219
902
2,032
20,165
29,347
778,430
26
Table 9. BY00 Eagle Network Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.118
0.121
0.277
0.295
0.064
0.056
0.206
0.827
0.264
0.835
0.703
0.376
0.435
0.517
0.187
Lower
95% C.L.
($1,000)
6,996
5,980
44
26
22,460
13,961
167
354
15
92
3,692
Cost
($1,000)
9,104
7,832
96
62
25,659
15,674
280
11
733
47
50
56
628
45
5,821
66,098
Upper
95% C.L.
($1,000)
11,212
9,683
148
98
28,857
17,386
393
30
1,112
124
119
98
1,164
91
7,949
27
Table 10. BY00 Western Network Costs and Confidence Intervals
Mail Category
1C Single-Piece Letters
1C Presort Letters
1C Single-Piece Cards
1C Presort Cards
Priority Mail
Express Mail
Periodicals
Standard Mail ECR
Standard Mail Regular
Parcel Post
Bound Printed Matter
Media Mail
US Postal Service
Free Mail
International Mail
Total
CV
0.319
0.165
0.354
0.910
0.087
0.109
0.443
0.946
0.490
1.000
1.000
1.000
0.738
0.000
0.342
Lower
95% C.L.
($1,000)
241
1,278
6
4,821
877
14
12
184
Cost
($1,000)
641
1,890
19
17
5,813
1,114
103
5
288
17
1
2
5
558
10,475
Upper
95% C.L.
($1,000)
1,041
2,503
32
48
6,806
1,352
192
14
564
51
3
7
13
932